H20 | AMD Phenom II C3 | Dragon's Revenge | Ultimate Thread

[wuttz]

chew* i got my NB cooler!



i prolly need to use a hammer on this.
i've been pushing on it since this afternoon.

[theoldtimer]

I disabled the IGP today to test some crossfire with the 4870x2’s. I also wanted to see if the 8GB of mixed ram would cause any trouble with the raid0 when I clocked the system up a little more. I did need to lower the memory divider to 1:2 in order for the system to boot into windows.

The 3D06 score was low for the test speed but it ran the test several times at this speed and didn’t crash or drop to the desktop.
I don't think the Vantage score is all that bad considering the installed hardware and OS.

MSI 890GXM-G65 MB -OS=Windows 7 RC 64-bit.
965-C3 @4.5GHz
8GB total memory 2x2GB Corsair 1600 CL9+ 2x2GB OCZ 1600 CL6
Test drive =4xOCZ 30GB Vortex SSD drives in Raid0 Partitioned as 20/100GB, boot drive=80GB-IDE
GFX=2x4870x2’s in quadfire @780/970
CPU cooling =chilled water (windshield washer fluid) fluid temp=-10/-15C

3D06
http://service.futuremark.com/compare?3dm06=13588099




Vantage
http://service.futuremark.com/compare?3dmv=2045814

[chew*]

chew* i got my NB cooler!



i prolly need to use a hammer on this.
i've been pushing on it since this afternoon.

A press at an automotive shop makes short work of it

[Hedghawg71]

or a socket the same size as the copper insert ans one that is way bigger. Place it in a vice (Small socket/intel P.O.S./big socket) and press it out that way. It takes almost no effort at all. All that hammering business will warp that copper slug all out of proportion.

[Zeus]

That 06 result looks pretty depressing for the clocks, what CPU-NB were you running there?

I hope this won't be the norm for 890 boards.

Windows 7 isn't the best for 06 but i would have expected the score to be a thousand points higher though.

[demonkevy666]

That 06 result looks pretty depressing for the clocks, what CPU-NB were you running there?

I hope this won't be the norm for 890 boards.

Windows 7 isn't the best for 06 but i would have expected the score to be a thousand points higher though.

probably the 8gbs of ram slowing it down, looks to only be running 1300mhz with 8-8-8-24(30)

[Zeus]

I think it's down to several reasons: Windows 7, 10.3 driver, low memclock+slack timings and lowish CPU-NB clock.

All that combined does not draw a right picture for 3d mark 06 performance of the board but i honestly doubt it will be better than a good 790FX board for 3D.

[chew*]

That 06 result looks pretty depressing for the clocks, what CPU-NB were you running there?

I hope this won't be the norm for 890 boards.

Windows 7 isn't the best for 06 but i would have expected the score to be a thousand points higher though.

Drivers and yes 06 in win7 = big hit.

[theoldtimer]

I think it's down to several reasons: Windows 7, 10.3 driver, low memclock+slack timings and lowish CPU-NB clock.

All that combined does not draw a right picture for 3d mark 06 performance of the board but i honestly doubt it will be better than a good 790FX board for 3D.

That post in question was more of a 64-bit OS test of the MB and chipset than anything else. I wanted to test Performance Test 7 with 8GB of ram but I don’t own four matching sticks of ram so I installed what I had on hand. Then I thought I would see if this combination would run every test in my list and it did, repeatedly. The 06 and Vantage tests were the only ones I posted. That combination of ram sticks didn’t want any timings below 8 and forget about the 1:4 divider as the system wouldn’t boot.
I think this MB could use some bios work because it doesn’t like the 1:4 divider and it’s sometimes a crap shoot as to whether the system will reboot after a simple setting change in the bios. I don’t know how much the pci-e lanes @x8 x8 (when two GFX cards are installed) effect the 3DMark06 score but Vantage seems ok. From my testing on the 4870x2’s with AMD and Intel MBs IMHO the 9.12 hot fix drivers are the best (at least for the 4870x2’s.) If I remember correctly the 9.12’s wouldn’t install when using the internal 4290 GFX chip so I’m on the 10.3’s (for that reason). I may remove the 10.3’s and try to install the 9.12 hot fix drivers again to see it they help the 06 score?

I did remove the 8GB of memory and install the 4GB Corsair GT’s with better timings but the other setting were mostly the same and still on the 64-bit OS.

The 06 score gained a bit the link is below.
http://service.futuremark.com/compare?3dm06=13592771

The tests below were done using different bios settings than during the 64-bit tests but the CPU clock speed was very close. As you can see there isn’t that much difference between the 64 and the 32-bit OS scores in 06 and Vantage.
Maybe I should put it this way based on the test settings and hardware used there wasn’t much difference between the scores of the 64/32-bit OS’s.

3D06 in Vista 32-bit 10.3 cats
http://service.futuremark.com/compare?3dm06=13594100


Vantage in Vista 32-bit 10.3 cats
http://service.futuremark.com/compare?3dmv=2048332

[Zeus]

Thanks for that test old, that looks a lot better. Still not too great due to that 10.3 driver but a lot better. Thanks for putting my mind at ease, 890 board is not scraped from my upgrade list now.

[Raiderman]

Is the 10.3 driver that bad?

[Zeus]

For 06, yes.

[theoldtimer]

Ok, I’ve been trying to find the reason for the low 3DMark06 when testing this MSI 890GXM-G65. I ran some single GFX card (4870x2) tests today. Eight total, four using PCI-e slot #1 and four using PCI-e slot #2. Three test were GFX tests and one full test per each PCI-e slot. Below is an excert from the manual. I read it as PCI-e slot #1 X16 when used alone, X8 when slot #2 is also used. Now PCI-e slot #2 is x8 when used as well as PCI-e slot #1 but what I don’t know is if slot #2 is x16 when used alone.

The whole point of this PCI-e test was to find out if maybe slot #2 was maxed @x8 and if so the when in crossfire with both slots running @x8 if this could be the cause for the low 3d06 score?.

[from the manual]
“Slots
2 PCI Express 2.0 x16 slots
when you install two expansion cards into both PCIE x16 slots, the PCI_E1 &
PCI_E2 lanes will run with x8 speed.
if you intend to install only one expansion card, please install it into PCI_E1 slot,
and the PCI_E1 lane will run with x16 speed
1 PCI Express x1 slot
1 PCI slot, supports 3.3V/ 5V PCI bus Interface”

When one card is installed CCC lists it as x16 when installed in slot #1, X8 when installed in slot #2 but when two cards are installed in crossfire CCC lists slot #1 as x8 and slot #2 x8. I included that CCC info in two of the screen shots below.

I have links for all of the GFX tests if you want to see them let me know. Anyway the average of the three GFX scores on PCI-e slot #1 was 12,497 and the average for the three GFX tests for slot #2 was 12,529. Maybe X8 is enough for a 4870X2?

The CPU was @ 4551MHz one 4870X2 @778/980MHz for All tests, the full 3d06 run was the fourth test in each set. Slot #2 was tested first.

PCI-e slot #2 first GFX test with CCC in SS



PCI-e slot #1 first GFX test with CCC in SS



PCI-e slot #2 full 3DMark06 test
http://service.futuremark.com/compare?3dm06=13601026



PCI-e slot #2 full 3DMark06 test
http://service.futuremark.com/compare?3dm06=13601545



3DMark06 PCI-e test quadfire 2x4870x2’s
http://service.futuremark.com/compare?3dm06=13607415

[64NOMIS]

Ok, I’ve been trying to find the reason for the low 3DMark06 when testing this MSI 890GXM-G65. I ran some single GFX card (4870x2) tests today. Eight total, four using PCI-e slot #1 and four using PCI-e slot #2. Three test were GFX tests and one full test per each PCI-e slot. Below is an excert from the manual. I read it as PCI-e slot #1 X16 when used alone, X8 when slot #2 is also used. Now PCI-e slot #2 is x8 when used as well as PCI-e slot #1 but what I don’t know is if slot #2 is x16 when used alone.

The whole point of this PCI-e test was to find out if maybe slot #2 was maxed @x8 and if so the when in crossfire with both slots running @x8 if this could be the cause for the low 3d06 score?.

Now that's what I'm talking about.

Sure looks like slot 2 is hard wired X8 and that it doesn't matter one bit even when the CPU is OC'd. Now go do the same thing with the Heaven benchmark using DX 10 setting....

[64NOMIS]

chew* i got my NB cooler!



i prolly need to use a hammer on this.
i've been pushing on it since this afternoon.

I want that cooler core alive...no disintegrations!

[64NOMIS]

Sidebar question, I have a drive that won't recognize consistently but will upon reboot; looks like the drive isn't powering up fast enough and isn't getting recognized by BIOS. Is there a way to delay IDE detection? Gigabyte 790FXTA-UD5P.

[Lastviking]

Sidebar question, I have a drive that won't recognize consistently but will upon reboot; looks like the drive isn't powering up fast enough and isn't getting recognized by BIOS. Is there a way to delay IDE detection? Gigabyte 790FXTA-UD5P.

Cant find anything in bios to delay it

[chew*]

Sidebar question, I have a drive that won't recognize consistently but will upon reboot; looks like the drive isn't powering up fast enough and isn't getting recognized by BIOS. Is there a way to delay IDE detection? Gigabyte 790FXTA-UD5P.

Maybe drive needs firmware update?

[Gener_AL (UK)]

will pressing pause on post not give time for drive to spin up?
or maybe setting some other bios options to slow down post?

[chew*]

Yah you may have quick boot enabled.....disable it.

[PatRaceTin]

i think 965 C3 is better than my 955 C3

it require 1.5++ Vcore to bench @ only 4.0-4.1 Ghz

[charged3800z24]

Yah you may have quick boot enabled.....disable it.

I was going to say the same thing. Plus, what drive is it? Did you say it was an actual IDE?

[theoldtimer]

I added a quadfire 3DMark06 test to the above set. I ran out of time yesterday so I didn’t get it in that set. The test is @ the same system clocks.

Now that's what I'm talking about.

Sure looks like slot 2 is hard wired X8 and that it doesn't matter one bit even when the CPU is OC'd. Now go do the same thing with the Heaven benchmark using DX 10 setting....

Thanks Simon, below is the Heaven Benchmark 2.0 pci-e tests x16, x8 and x8 + x8.

Again as earlier I ran three tests of each PCI-e slot #1 x16 and #2 x8 using 1x4870x2 GFX card. I averaged the results of each three run set to find the test results for each PCI-e slot. The screen shots are of the last test or third test of each three test set. I also posted one screen shot (bottom of post)of a test ran in quadfire (2x4870x2’s).

The CPU @4551MHz 1x4870x2 @778/980MHz Vista 32-bit

PCI-e slot #1 average of three tests
FPS=103.6
Score=2610
Min FPS=52.6
Max FPS=209.4

PCI-e slot #2 average of three tests
FPS=102.7
Score=2586
Min FPS=51.4
Max FPS=207.9


PCI-e test slot #1 x16 third test SS with CCC




PCI-e test slot #2 x8 third test SS with CCC



PCI-e test both slot #1 and #2 both @ x8 quadfire 2x4870x2’s

[wuttz]

very nice testing with the pci-e slots/bandwidth oldtimer!
so it looks like there is no "bottleneck" even with just using x8 mode.

[64NOMIS]

I added a quadfire 3DMark06 test to the above set. I ran out of time yesterday so I didn’t get it in that set. The test is @ the same system clocks.

Thanks Simon, below is the Heaven Benchmark 2.0 pci-e tests x16, x8 and x8 + x8.

Again as earlier I ran three tests of each PCI-e slot #1 x16 and #2 x8 using 1x4870x2 GFX card. I averaged the results of each three run set to find the test results for each PCI-e slot. The screen shots are of the last test or third test of each three test set. I also posted one screen shot (bottom of post)of a test ran in quadfire (2x4870x2’s).

Sig Worthy. Xtremely Helpful.